http://informahealthcare.

com/pdr
ISSN: 1751-8423 (print), 1751-8431 (electronic)

Dev Neurorehabil, 2014; 17(2): 137–146

! 2014 Informa UK Ltd. DOI: 10.3109/17518423.2014.906002

ORIGINAL ARTICLE

Orthographically sensitive treatment for dysprosody in children with

Childhood Apraxia of Speech using ReST intervention
Patricia McCabe, Anita G. Macdonald-D’Silva, Lauren J. van Rees, Kirrie J. Ballard, & Joanne Arciuli

Discipline of Speech Pathology, Faculty of Health Sciences, The University of Sydney, Sydney, NSW, Australia

Abstract Keywords
Objective: Impaired prosody is a core diagnostic feature of Childhood Apraxia of Speech (CAS) Childhood apraxia of speech, lexical stress,
but there is limited evidence of effective prosodic intervention. This study reports the efficacy orthography, prosody, dyspraxia,
of the ReST intervention used in conjunction with bisyllabic pseudo word stimuli containing intervention
orthographic cues that are strongly associated with either strong-weak or weak-strong patterns
of lexical stress. Methods: Using a single case AB design with one follow-up and replication, four History
children with CAS received treatment of four one-hour sessions per week for three weeks.
Sessions contained 100 randomized trials of pseudo word treatment stimuli. Baseline measures Received 5 January 2014
were taken of treated and untreated behaviors; retention was measured at one day and four Revised 12 March 2014
weeks post-treatment. Results: Children’s production of lexical stress improved from pre to Accepted 13 March 2014
post-treatment. Treatment effects and maintenance varied among participants. Conclusions: Published online 2 April 2014
This study provides support for the treatment of prosodic deficits in CAS.

Introduction weaker stress. Thus younger English speaking children having

more success with SW words than WS ones [7, 9].
Dysprosody in Childhood Apraxia of Speech
High levels of variability and inconsistency in the
Atypical prosody is thought to be a key feature of Childhood production of stress are noted in children with CAS when
Apraxia of Speech (CAS) [1, 2]. The production of lexical compared with other children or adults with speech impair-
stress, in particular, is an area of weakness in CAS. While ments [9–11]. Atypical prosody has been said to differentiate
individuals with CAS have been shown to mark stress in children with CAS from other children with speech delay or
similar ways to younger but typically developing children [3], phonological disorder and higher levels of dysprosody occur
they appear to use longer word and segment durations than in younger children with CAS than children with other speech
their typically developing peers and tend not shorten duration disorders [12]. In their repetition of both SW and WS
of vowels in unstressed initial syllables [4]. This paper reports nonwords, children with CAS are perceived to match lexical
a study designed to examine use of Rapid Syllable Transition stress patterns less frequently than phonologically disordered
Training (ReST), in conjunction with orthographically biased children, even when no differences are detected on acoustic
stimuli, to treat disordered prosody in children with CAS. measures [13]. Munson and colleagues [14] also noted that
Prosody is realized acoustically through the manipulation children with CAS mark SW words with pitch and loudness 14
20
of vowel duration (ms), vocal intensity (dB), and vocal pitch contrasts and WS words with duration contrasts while other
(i.e. fundamental frequency or F0 in Hz) [5]. In languages children do not make this distinction. Excessive, equal or
such as English, stressed syllables carry the longest vowel misplaced stress has also been reported to occur more often in
duration and higher peak F0 and intensity compared with children with CAS when compared with other pediatric
other syllables within the word. The dominant lexical stress speech populations and adults with acquired Apraxia of
pattern for English nouns is strong-weak (SW) where stress is Speech [13, 15]. These various stress differences between
placed on the first syllable of words and subsequent syllables typically developing children and children with CAS may lead
are shorter, softer or lower in pitch. English speaking children to the perceived significant difficulties in both understanding
who are typically developing show a reliance on this pattern and treating the speech of these children [16].
in their speech when learning other lexical contrasts such as Prosodic impairments impact significantly on overall
weak-strong (WS) where the stress is placed on the second speech intelligibility although relatively few studies consider
syllable of a word [6–8] and the preceding syllable carries prosody when discussing intelligibility. A review of the
literature, looking at the interaction of prosody and speech
intelligibility in disorders of speech such as dysarthria,
Correspondence: Dr. Patricia McCabe, Speech Pathology, Faculty of
Health Sciences, University of Sydney, PO Box 170, Lidcombe, Sydney, emphasized the impact of prosody on overall speech intelli-
NSW 1825, Australia. E-mail: [email protected] gibility [16]. For example, in people with dysarthria,
138 P. McCabe et al. Dev Neurorehabil, 2014; 17(2): 137–146

judgments of both impaired prosody and measures of with an orthographically biased pseudo word such as
articulatory accuracy correlate highly with perceived speech ‘‘bedoon’’ the children tended to assign a WS pattern
intelligibility [17]. Thus, prosodic impairment can affect during reading aloud. Older children were found to be more
understanding of speech at similar levels to the loss of sensitive to these cues than younger children and this
intelligibility caused by the segmental speech impairment. developmental trajectory was further explored using compu-
Additionally, we now know that, in English, both supraseg- tational modeling. It was concluded that sensitivity to these
mental and segmental information assist lexical access during probabilistic orthographic cues is most likely the result of
the recognition of spoken words [18–20]. Incorrect placement implicit learning that grows over time with increasing
of stress has been shown to impede lexical access [21]. exposure to a broader range of written materials.
Van Rees and colleagues recently drew on Arciuli’s work
in their use of orthographic stimuli containing probabilistic
Treatment for CAS
cues to lexical stress in a study examining explicit training of
Most existing interventions for CAS focus primarily on stress assignment in typically developing children [29]. Using
articulatory parameters and little consideration has been given the approach outlined by Ballard et al., Van Rees et al.
to treatment of prosodic impairments and particularly to reported using explicit instruction to attend to the relative
measurement of resultant improved prosody in this disorder. length of each syllable as a part of the training and reported
With the exception of the team led by Strand [22–24] who perceptual judgment of stress production as the primary
described Dynamic Temporal and Tactile Cueing (DTTC) outcome measure. Typically developing children were ran-
intervention and the ReST intervention [25], no studies report domly allocated to either learn the stimuli (the experimental
treatment for prosody specifically. group) or be in a control group. Children in the experimental
DTTC intervention includes instruction for varying pros- group learnt to say the pseudo word stimuli in minimum time
ody within treatment trials while focusing on the articulatory while children in the control group did not. Van Rees et al.
accuracy of target words. In DTTC, children imitate words or concluded that this approach could be applied in future
phrases with varied intonation patterns following a model studies, in an attempt to remediate dysprosody.
provided by the clinician. This variation in prosody is not Indeed, no previous prosodic intervention for CAS has
variation in lexical stress but rather variation in sentence been taken into account the biases toward different lexical
stress. To date DTTC studies have not reported measurement stress patterns that may arise with various orthographic
of prosody to document changes to prosodic accuracy as a stimuli. It seems possible that the use of orthographic stimuli
direct result of this approach. may either enhance or hinder the treatment depending on
By contrast, the ReST treatment specifically targets the whether the stimuli contain spelling patterns that are in line
ability to control relative durations in the production of SW with the stress patterns that the child is being asked to
and WS pseudo words while simultaneously producing produce. This applies to studies using either real words or
accurate speech segments at an age appropriate speech rate. pseudo words. However, pseudo words are especially
Thus, ReST involves targeting production of lexical stress. appropriate for use in treatment because they are not
Ballard and colleagues [25] reported that treatment stimuli influenced by previously learned motor plans, or differences
consisting of cloze sentences with target three syllable pseudo in frequency or familiarity across participants [30]. To
words as the final word in the sentence were modeled by the illustrate the role of orthographic bias, if a child was being
clinician in the pre-practice phase where the children imitated presented with a written pseudo word such as ‘‘bedoon’’ and
the clinician. The stimuli were then read by the children was being asked to produce SW stress, that child could,
without a model in the practice phase [25]. All three potentially, struggle to overcome a preference for assigning
participants in that multiple baseline design study improved WS stress. Presentation of the written pseudo word
prosodic control including duration, loudness, and/or pitch ‘‘bedoon’’ in conjunction with instruction on the production
contrasts for both SW and WS stress patterns and generaliza- of a WS pattern might be likely to facilitate success in stress
tion was noted to untreated but similar stimuli. production.
A separate body of work by Arciuli and colleagues has Here, we extend the ReST intervention for CAS [25] by
demonstrated that there are orthographic markers for lexical using Van Rees et al.’s [29] pseudo word stimuli. This is the
stress present in the spelling patterns of English words [8, 26– first prosody intervention study to use such treatment stimuli.
28]. Analyses of both child and adult corpora have revealed Hypotheses for this study were:
that the beginnings and endings of bisyllabic words are (1) Children with CAS will show significant improvement in
indicators of stress position. Most recently, Arciuli et al. [8] the ability to contrast SW and WS lexical stress patterns
examined a database of almost 20 000 disyllabic words from in both treated and similar untreated exemplars of biased
children’s reading materials in an effort to identify probabil- two-syllable pseudo words.
istic orthographic cues to lexical stress. To illustrate the kinds (2) The effects of treatment will be retained up to four weeks
of findings they reported, the analyses revealed that most post-treatment and will generalize to similar untrained
bisyllabic words beginning with ‘‘be’’ and most bisyllabic stimuli.
words ending with ‘‘oon’’ have WS stress. In behavioural (3) Treatment effects will generalise to connected speech as
testing of 186 typically developing children aged 5–12 years a measure of ecological validity.
using carefully constructed pseudo words that contained these (4) Treatment effects will not generalize to improved recep-
probabilistic cues, Arciuli and colleagues demonstrated that tive vocabulary skills, demonstrating experimental
children are sensitive to these cues. That is, when presented control.
DOI: 10.3109/17518423.2014.906002 Prosodic treatment of CAS 139

Method child, transcribed phonemically and entered into the

Computerized Profiling program to calculate speech accuracy
Participants
measures [36]. The ASHA core CAS features [37] were used
This research was approved by The University of Sydney by two independent and experienced speech pathologists to
Human Research Ethics Committee (number 11317). confirm CAS diagnosis from the single word, inconsistency
Participants were recruited through electronic advertisements, and connected speech samples reported in Table I. These
the university clinic and community speech language path- features included inconsistent sound errors on repetitions of
ologists. Four children met the inclusion criteria: history and syllables or words (as scored on the DEAP inconsistency
current diagnosis of CAS; normal receptive language skills subtest [38]); the presence of poor control of coarticulatory
(Clinical Evaluation of Language Fundamentals, 4th edition, transitions between sounds and syllables; and inappropriate
Australian Standardization CELF-4 [31]); Peabody Picture prosody, particularly of lexical or phrasal stress. Table I
Vocabulary Test, PPVT [32]; normal hearing acuity [33]; no contains a summary of test results for both eligibility and
known other developmental or genetic diagnosis; and English description of participants.
as their first language and at least one parent with English as
their first language. Oral motor examination [34] revealed no
abnormalities in orofacial structure, muscle strength, muscle Reading ability
tone, or reflexes for any participant. Connected speech As the treatment centered on visually presented stimuli,
samples were collected for each child, following McLeod’s reading ability was assessed using the Neale Analysis of
procedures [35], with at least 50 utterances collected for each Reading Ability, 3rd edition (NARA-3) [39] and the Word

Table I. Pre-treatment assessment battery results for P1, P2, P3, and P4.

P1 P2 P3 P4
(male, 8 y; 6 m) (male, 6 y; 7 m) (male, 6 y;6 m) (male, 5 y; 5 m)
Test (used for) Std CI Std CI Std CI Std CI
Peabody Picture Vocabulary Test – 4th edition Form A 90 83–97 90 83–97 91 84–98 117 109–124
(receptive vocabulary)
Clinical Evaluation of Language Fundamentals – 4th edition (receptive & expressive language ability)
Core Language Score 57* 49–55 81 74–88 76 69–83 89 84–94
Receptive Language Index 103 93–113 96 87–105 103 94–112 105 97–103
Expressive Language Index 53* 44–62 76* 68–84 74* 66–82 80* 74–86
Test of Auditory Processing – 3rd edition
Word Discrimination 5* 3.7–6.3 8 6.4–9.6 8 6.4–9.6 9 7.3–11.6
(auditory discrimination of real words)
Word Memory (verbal memory) 6* 4.6–7.4 10 8.6–11.4 6* 4.6–7.4 12 10.7–13.3
Woodcock Reading Mastery Test – Revised
Basic skills cluster 88 86–89 127 126–129 94 88–100 104 101–107
Word identification (reading level) 90 88–91 129 127–130 98 90–102 104 99–108
Word attack (nonword reading) 89 87–91 121 119–122 81* 28–103 105 94–110
Lower Case Letters checklist percent 89 – 94 – 48 – 59 –
(letter name knowledge)
Comprehensive Test of Phonological Processing Std ±SEM Std ±SEM Std ±SEM Std ±SEM

Phonological Awareness Composite score 64* 59–69 106 103–109 85 82–88 85 82–88
Phonological Memory Composite score 70* 64–76 91 84–98 88 81–95 88 82–94
Memory For Digits (verbal memory) 6* 4–8 10 9–11 7 6–8 8 7–9
Non-word Repetition (nonword repetition) 4* 3–5 7 6–8 9 8–10 8 7–9
Rapid Naming Composite score 91 86–96 94 89–99 118 113–123 136 131–141
Neale Analysis of Reading Ability – 3rd edition Reading Reading
(reading level) age ±SEM age ±SEM

Accuracy 6.2 – 56.0–6.8 – 7.0 – 6.5–7.7 –

Comprehension 6.11 – 6.3–7.3 – 6.9 – 6.3–7.5 –
Rate 7.4 – 6.8–8.0 – 12.9 – 11.7–413 –
Connected speech % % % %

Percent Vowels Correct (PVC) 74 – 77 – 90 – 85 –

Percent Consonants Correct (PCC) 81 – 95 – 70 – 60 –
Stress pattern match percent (core CAS feature) 46 – 53 – 79 – 63 –
Inconsistency percent (Dodd, 2002) (core CAS feature) 20 24 36 40

Standardized tests scores (Std) and 95% confidence intervals (CI) or Standard Errors of Measurement (SEM) are reported as appropriate for
assessments.
– Refers to assessments and/or subtests not administered due to age/ability or where range scores not available.
* Indicates a score 1 or more standard deviations below the mean.
% indicates a percentage score.
140 P. McCabe et al. Dev Neurorehabil, 2014; 17(2): 137–146

Identification subtest of the Woodcock Reading Mastery Test- Audition 1.0 [44] with a sampling rate of 44 100 Hz [45].
Revised (WRMT-R) [40]. These were used to determine each A 5 cm microphone-to-mouth distance was used.
participant’s reading level. Nonword reading was assessed
using the WRMT-R Word Attack subtest. Nonword speech Treatment procedures
repetition was assessed using the Comprehensive Test of Following Ballard et al. [25] and van Rees et al. [29],
Phonological Processing (CTOPP) [41]. Participants P1 and treatment sessions were of 60 min duration, four days per
P2 were able to read passages from the NARA-3, identify week for three weeks (total 12 sessions per participant).
words in the Word Identification subtest, and decode Participants and their parents were instructed to undertake no
nonwords in the Word Attack subtest of the WRMT-R and additional practice of treatment targets outside of these
so judged capable of independently reading the stimuli sessions. Neither children nor their parents were advised of
presented in treatment. Participants P3 and P4 were unable the hypotheses; however, it was not possible to blind
to identify words or decode nonwords and so were not clinicians to the research hypotheses. Clinicians received a
administered the NARA-3 and were supported in treatment treatment manual and explicit instruction, training, and
with clinician spoken modeling of stimuli for direct imitation. demonstration of treatment procedures prior to commencing
sessions [46]. All sessions were video-recorded for later
Experimental design scoring of reliability on dependent measures. Consistent with
A single case AB design with one follow-up with replication a PML approach [47], sessions consisted of a pre-practice
was employed to test treatment-related changes from pre- to component (10–20 min) and a practice component (40–
post-treatment in the four children. All participants completed 50 min), as follows.
three baseline sessions, followed by 12 one-hour treatment
sessions over three weeks. Experimental probes were admin- Pre-practice
istered after every fourth session, on a separate day to During pre-practice, clinicians first presented a randomly
treatment, and a single probe was completed at four weeks selected stimulus item and asked the child to identify whether
post-treatment. No feedback on production accuracy was it had a SW (long-short) or a WS (short-long) pattern. If
provided during these probes, which tested for acquisition and necessary, the clinician corrected the response. Next, the
retention of treated behaviors and generalization of treatment clinician modeled the pseudo word for imitation and provided
effects to untrained stimuli. In addition, a connected speech 100% knowledge of performance (KP) feedback on stress
sample was collected at each testing point to evaluate pattern (e.g., ‘‘Try to make the first part even shorter’’) to
generalization of treatment effects in a more ecologically shape further attempts. All models were presented with
valid context. Receptive vocabulary skills were tested pre- sentence intonation pattern. Other cues were offered as
and post-treatment as a measure of experimental control as appropriate such as hand tapping to cue rhythm and target
this treatment that uses pseudo words was not expected to length of syllables. This feedback and shaping cues allowed
accelerate the development of receptive vocabulary. participants to experience successful production. Although
knowledge of results (KR) feedback (i.e., indicating only the
Stimulus materials and equipment correctness of a response) was given on segmental errors, no
Baseline/experimental probe and treatment stimuli consisted specific articulatory or phonetic placement instructions were
of the same 30 bisyllabic pseudo words used in van Rees’ study provided. Participants were moved to the practice phase when
[29] which were orthographically biased to either a strong- they had produced five consecutive correct trials, including
weak (SW; e.g., mandan) or a weak-strong (WS; e.g., bedoon) both SW and WS tokens.
stress pattern, based on the work of Arciuli and colleagues
[8,26] (Appendix). Pseudo words were used as these cannot be Practice
influenced by learned motor or linguistic plans, which may be The 19 orthographic treatment stimuli were presented in
related to frequency of use or familiarity, that are present when random order with at least 100 trials per session, for a minimum
using real words [25,30]. Speech interventions using pseudo target of 1200 trials per child. Participants were instructed
words have been shown to stimulate generalization of either to read the word aloud or to repeat the word after the
treatment effects to real words [42]. clinician dependent on the presentation method assigned from
Nineteen treatment pseudo words were selected randomly pretesting. KR feedback was given (i.e. ‘‘good’’/"not good’’)
from the list of 30 with the remaining 11 serving as probes to for combined prosodic and segmental accuracy on 50% of all
assess acquisition of skills to untreated related stimuli. All trials, fading from high to low frequency across the session,
pseudo words were displayed orthographically on flash cards with a delay between response and provision of feedback of
in 72 point Times New Roman font on 13.2  4.7 cm index 3–5 s. No KP feedback was provided. Mastery was set at 80%
cards with a different picture of an alien accompanying each correct over three consecutive sessions. It should be noted that
word [43]. Pictures were provided for each pseudo word to typically developing children achieve this mastery on an
make them visually more interesting thereby enhancing the average within 3–4 sessions [29].
distinctiveness of each of the pseudo words.
All baselines and experimental probes were digitally
Variation to practice procedure
recorded in a sound-treated booth using a Layla 24/96
Multitrack Recording System with C420III – PP Micro-Mic II Across sessions one to five, it was noted that all participants
head-mounted microphone (AKG Acoustics) and Adobe had consistent idiosyncratic, incorrect productions for some
DOI: 10.3109/17518423.2014.906002 Prosodic treatment of CAS 141

pseudo words. For example, P1 produced ‘‘mandan’’ as an accurate production, the pseudo word had to be correct for
/m"nd|d/ (a real word) instead of /mænd|n/ (a pseudo word). both stress and segments. Intra-rater agreement on phonemic
Although the lexical stress pattern was correct, we decided to transcription ranged from 95 to 97% across participants, and
eliminate these idiosyncratic productions with short blocks of inter-rater agreement from 83 to 90%.
trials. This short-block procedure was triggered when the Reliability of perceptual intra-rater and inter-rater agree-
error occurred on a trial that had, a priori, been randomly ment was also calculated on a random 14% of baseline and
selected to receive KR feedback. That is, a participant was experimental probe data for each participant. The intra-rater
given KR feedback on prosodic and segmental accuracy on agreement was 93% and the inter-rater agreement was 88%.
the first attempt at the stimulus item; that is, if the production Reliability of treatment provision (fidelity) was calculated
was correct the child received KR feedback and was presented on compliance with the protocol during the practice phase.
with the next trial, but if the idiosyncratic production occurred A random block of 20 stimulus–response pairs was examined
s/he received modified KR feedback (e.g. ‘‘good pattern, for each session and scored for treatment fidelity. Trials were
wrong sounds’’) and was prompted to try again. Up to three considered to have fidelity if all components were correct
extra attempts were allowed with KR feedback on each including: (1) correct presentation of stimulus, including
attempt. The block trials were interwoven with, but in correct phonemic and prosodic spoken model for P3 and P4;
addition to, the 100 trials per session. (2) provision of delayed feedback; (3) provision of KR
feedback only; and finally (4) provision of feedback only on
Dependent measures the pre-randomized items. Fidelity was as follows P1 mean
78% SD 13; P2 mean 76% SD 15, P3 mean 83% SD 16, and
Productions of pseudo word stimuli during baselines and p4 mean 70% SD 17. For the whole group, fidelity was
experimental probes and during treatment sessions were therefore mean 75% SD 16. Most lapses in fidelity were due
judged perceptually as correct or incorrect based on the three to errors in delaying feedback (both substantially shorter and
measures – (1) stress pattern correct; (2) all segments correct; longer delays than the desired 3 s) and occurred more
and (3) simultaneously correct stress and segments; the latter frequently in earlier sessions than in later ones.
having been shown in studies by Ballard et al. [25] and
McCabe et al. [48] to be harder than either separately. For Results
stress pattern, a response was judged as correct if the vowel in
the strong syllable was perceived as a full vowel (i.e. not Performance on treated pseudo words during practice
reduced) and the vowel in the weak syllable was perceived as During treatment sessions, all participants demonstrated
a schwa. For segmental accuracy, responses were further improved prosodic and segmental accuracy of SW and WS
subcategorized as percent vowels correct (PVC) and percent bisyllabic pseudo words, although none reached the mastery
consonants correct (PCC) as calculated in the PROPH module criterion of 80% correct over three consecutive sessions
of Computerized Profiling [36]. (Supplementary figures). P1 initially demonstrated both
Connected speech samples were collected at each probe prosodic and segmental accuracy below 10% correct and
session to examine for generalisation of treatment effects to a steadily improved to 66% on both measures in the final
functional speaking task. A minimum of 50 utterances were session. P2 initially had prosodic accuracy below 30% and
phonemically transcribed and entered into Computerized segmental accuracy below 10%; with prosodic accuracy
Profiling [36]. reaching 82% correct in the final session and segmental
accuracy fluctuating between 39 and 59% over the final three
Data analysis sessions. Initially, the P3’s prosodic accuracy was 15% and
Perceptual treated and untreated probe item data were the segmental accuracy was about 50%. Both measures
graphed for visual analysis and treatment effects analyzed improved over time, reaching a peak of 75% in the second last
using percent nonoverlapping data (PND). Scruggs and session. P4 initially demonstrated prosodic accuracy around
Mastropieri [49] (p. 224) suggested that a value of over 30% and segmental accuracy around 50%. Prosodic accuracy
70% datapoints nonoverlapping across baseline and treatment increased to an average of 58% in the final three sessions and
phases of the study for treated or untreated stimuli suggests segmental accuracy increased to an average of 79%.
a clear treatment or generalization effect, respectively, while
scores between 50 and 70 have questionable effect and scores Treatment, retention, and generalization effects
under 50 should be interpreted as no demonstration of treat- As this was a within-subject experimental design study,
ment effect. Sub-analysis of SW and WS accuracy is reported results are interpreted for each participant individually.
as raw scores for each participant. Figure 1 shows the perceptual measures of treated and
untreated probe items for each participant and Online
Reliability Supplementary Figures 2–5 show the analysis of SW and
Inter- and intra-rater reliability on dependent measures was WS items for each participant.
calculated for 20% of trials for each treatment session for all
Participant 1
participants; that is, inter-rater agreement on judgment of
response accuracy during treatment and intra- and inter- rater PND across three probes and retention on treated items was
agreement on phonemic transcription of responses. As per 75 and on untreated items was 25. With regard to SW and WS
Ballard et al. [25] and Van Rees et al. [29], to be recorded as stimuli, P1 improved from 0% accuracy to 50% correct for
142 P. McCabe et al. Dev Neurorehabil, 2014; 17(2): 137–146

P1: Percent perceptual accuracy P2: Percent perceptual accuracy

100% 100%
90% 90%
80% 80%
70% 70% PND Treated 100%
60% PND Treated 75% 60%
50% Untreated 25% 50% Untreated 25%
40% 40%
30% 30%
20% 20%
10% 10%
0% 0%

P1 Treated P1 Untreated P2 Treated P2 Untreated

P3: Percent perceptual accuracy P4: Percent perceptual accuracy

100% 1
90% 0.9
80% 0.8
70% 0.7 PND Treated 100%
60%
PND Treated 75% 0.6
50% Untreated 75% 0.5 Untreated 0%
40% 0.4
30% 0.3
20% 0.2
10% 0.1
0% 0

P3 Treated P3 Untreated P4 Treated P4 Untreated

Figure 1. Overall perceptual performance for each participant on treated and untreated probe items across three periods (baseline, in treatment probes,
4 week follow up). Percent is number of probe items child was perceived to have both correct prosody and articulation as a percent of number of trials
respectively on treated and untreated items. PND – Percent non overlapping data.

prosody on SW pseudo words but only 10% combined stimuli, P3 showed improvement on SW pseudo words during
accuracy immediately post-treatment (i.e. the third and final treatment primarily for segmental and combined accuracy
probe during the treatment phase), with stable retention (from 12% to 50%), with gains remaining stable in
four weeks post-treatment. WS pseudo words were at retention. Prosodic accuracy was reasonably stable around
0% accuracy during baseline and, while prosodic and 60–67% in baseline and treatment phases but increased to
segmental accuracy each improved to 15–20% post-treatment, 88% at retention, despite receiving no intervention during this
combined accuracy remained at 0%. Interestingly, this four-week period. WS pseudo words showed a trend to
participant reached 60% combined accuracy, with 90% declining accuracy during baseline but ended above baseline
prosodic accuracy at retention, despite receiving no interven- levels at post-treatment with 55–60% correct prosody and
tion during this four-week period. segments, respectively, and 40% combined accuracy. Prosody
continued to improve during retention, to 68%, but segmental
Participant 2 and combined accuracy deteriorated to baseline levels.
PND across three probes and retention on treated items was
100 and on untreated items was 25. With regard to SW and Participant 4
WS items, P2 improved from 0 to 10% accuracy on prosody
and combined accuracy for SW pseudo words in baseline to PND across three within treatment probes and retention on
between 35 and 50% correct at post-treatment. There was treated items was 100 and on untreated items was 0.
some loss of skills at retention, but performance remained Regarding SW and WS stimuli, data from baseline 1 were
above baseline levels on treated items. WS pseudo words unavailable due to failure of recording equipment. Over time,
showed clear improvement from 0 to 5% accuracy on all three P4 improved on all three perceptual measures for SW pseudo
perceptual measures in baseline to 40–45% accuracy at post- words, but accuracy appeared to improve in baseline and
treatment. Again, performance deteriorated during the reten- continue on a similar slope through treatment and retention
tion phase but remained above baseline levels. phases, not supporting a specific intervention effect. For WS
pseudo words, prosodic accuracy was high in baseline
(70%) then dropped during treatment to stabilize around
Participant 3
54% at post-treatment and retention. Segmental and combined
PND across treatment and retention on treated items was 75 accuracy showed no evidence of a treatment effect, fluctuat-
and on untreated items was 75. Regarding SW versus WS ing between 5 and 35% correct over the three phases.
DOI: 10.3109/17518423.2014.906002 Prosodic treatment of CAS 143

Generalization to connected speech Table II. Percent vowels correct (PVC), percent consonants correct
(PCC), and percent stress pattern matches (stress %) during connected
Changes in PVC, PCC and stress pattern matches obtained speech at pre-treatment (Pre), post-treatment (Post) and retention (Ret)
from connected speech samples at pre-treatment (baseline 3), for each participant..
post-treatment (1 day immediate post) and retention (4 weeks
P1 P2 P3 P4
post) probe time points are presented in Table II. All
participants increased in their ability to produce correct Pre Post Ret Pre Post Ret Pre Post Ret Pre Post Ret
stress patterns during connected speech from pre-treatment to PVC 74 81 80 78 87 87 90 88 92 85 84 91
the retention probe. P1, P2, and P3 increased in stress matches PCC 81 80 54 95 95 88 70 86 74 60 70 66
by 19–23 percentage points from pre-treatment to retention, Stress % 46 43 70 53 81 76 79 77 85 64 68 83
while P3 increased by 6 percentage points.
Segmental accuracy increased in connected speech for
some participants. Small positive changes in PVC over time Table III. Post-treatment assessment results for P1, P2, P3, and P4.
were noted for P1, P2, and P4, with these participants
showing increasing PVC values over time by 6–10 percentage P1 P2 P3 P4
points. For PCC, P1 and P2 showed deterioration in accuracy
Test Std CI Std CI Std CI Std CI
from pre-treatment to retention (27 and 7 points, respect-
ively). P3 and P4 showed small increases in PCC of 4 and 6 Peabody Picture 84 78–91 96 89–103 90 83–97 119 111–125
Vocabulary
percentage points, respectively. Test – 4th edition
Form B

Experimental control Standardized tests scores (Std) and 95% confidence intervals (CI) are
reported, % indicates a percentage score.
Results of post-treatment language assessment are reported in
Table III. While small improvements were noted for receptive
problems reported by parents to be associated with regular
vocabulary scores on the PPVT [32], these were within the
home practice including finding time for daily practice [50]
95% confidence interval of the mean suggesting no systematic
and knowing that they are providing an accurate model for their
change related to the treatment, as expected.
child [51]. It should be noted, however, that reading ability and
difficulties decoding pseudo words may influence both stress
Discussion
assignment and segmental accuracy of spoken stimuli. While it
This study evaluated a treatment protocol targeting improved is difficult to control for this effect during treatment (i.e.,
prosody and segmental accuracy in children with CAS. determining if errors are the result of misreading and/or
Specifically, we examined the use of ReST intervention using misarticulation), it needs to be considered when determining if
bisyllabic pseudo word stimuli orthographically biased for children are going to be asked directly to read treatment
particular lexical stress patterns. We hypothesized that (1) stimuli. Gillon and colleagues have shown that children with
children would improve in their ability to produce lexical CAS may experience delays in reading ability [52,53] and,
stress in novel two-syllable pseudo words; (2) treatment therefore, clinicians should cautiously consider written-only
effects would be retained up to four weeks post-treatment; orthographically biased stimulus. This applies not only to
(3) treatment effects would generalise to connected speech; prosodic interventions but also to treatments targeting other
and (4) experimental control would be demonstrated. The characteristics of CAS such as segmental errors.
hypotheses were supported, however, the magnitude of the Where written stimuli are used, particular attention should
treatment effect, retention, and generalization varied across be paid to the nature of these stimuli as it seems possible that
participants. Based on perceptual judgments of prosodic, certain stimuli may enhance or hinder the treatment depend-
segmental and combined accuracies (the ability to produce ing on whether the stimuli contain spelling patterns that are in
both correct stress pattern and correct phonemes), all children line with the stress patterns that the child is being asked to
improved in their ability to produce both lexical stress produce. This applies to studies using either real words or
contrasts and segmental accuracy from the initial to final pseudo words. The combination of this study and that of Van
treatment sessions in treated words, thus a treatment effect Rees et al. [29] suggests that the use of appropriately selected
occurred. Participants also produced noticeable differences written stimuli which are orthographically biased can enhance
between SW and WS stimuli at more than one time point as production accuracy for children with CAS and those with
was expected with the focus of treatment being the vowel typical speech development.
length produced in the weak syllable with criterion set at Participants also made changes in segmental accuracy both
production of the schwa vowel. in therapy and in the generalization measure of connected
Both children who read the stimuli (P1 and P2) improved speech. This occurred even though only simple Knowledge
their production of treated words and the effects of treatment of Results [47] feedback on correct/incorrect production
generalized to untrained probe items and to connected speech. was provided on segmental accuracy. That is, participants
Thus, reading orthographically biased stimuli may be a were only told that sounds were correct or incorrect in the
functional approach to treatment of CAS and it may be that pre-practice phase and not in the practice phase and no direct
imitation of a clinician model, which is common practice in teaching on sound accuracy was provided at any stage. This
speech intervention, is not required by older children with finding echoes work with adults with acquired apraxia of
CAS. This finding may help ameliorate a number of the speech where both Mauszycki and Wambaugh [54] and
144 P. McCabe et al. Dev Neurorehabil, 2014; 17(2): 137–146

Brendel and Zeigler [55] reported training rate or rhythm assigned and all clinicians were required to demonstrate
alone also improved segmental accuracy. treatment fidelity and so delivered the protocol as intended.
All participants had difficulty synchronously producing A ceiling effect was observed for P4 with the bisyllabic
both correct stress patterns and correct segments. These stimuli in isolation being insufficiently challenging. Future
difficulties are reflected in the perceptual judgments of both studies should manage this effect through the provision of
treatment and probe data where most participants had poor more complex orthographically biased stimuli and the use of
performance on simultaneous production of the correct stress such stimuli in cloze sentences as per Ballard et al.’s
pattern whilst maintaining segmental accuracy. This inability approach [25].
to combine performance across speech elements may be These four single cases provide evidence that a treatment
hypothesized as causally linked to an increased load on the effect can be obtained using the current protocol. Additional
overall speech system. For example, P3 maintained only research is needed to further test this treatment approach in
prosodic accuracy after treatment and P4 was able to achieve a larger group of children with CAS using both imitation
accurate prosody alone on most occasions with segmental and spontaneous reading of the stimuli and to assess the
accuracy apparently sacrificed to achieve prosodic accuracy effects of orthographically biased treatment stimuli in a range
[56]. Stress pattern matches during connected speech of prosodic disorders. Further development of the protocol
increased while PCC values decreased for P1 and P2 also and validation with a wider range of children with CAS is also
suggesting a sacrifice of segmental accuracy to achieve required before we can be assured of the efficacy of this
prosodic accuracy. Such a trade-off between development of promising intervention.
lexical stress and segmental accuracy is seen in the develop-
ment of polysyllabic words in typically developing children Conclusions
with prosodic and/or segmental accuracy sacrificed at differ- Prosodic impairments can be treated in children with CAS
ent stages of development until an adult like production is using ReST intervention in conjunction with orthographically
realized [57]. Therefore, participants in this study may be biased stimuli. The use of stimuli orthographically biased to
following a normal developmental pattern in the development particular lexical stress patterns resulted in positive treatment
of lexical stress with positive changes to stress production outcomes for all children with maintenance and generalization
triggered as a result of treatment [6]. of treatment skills.

Generalization Acknowledgements
Generalization to non-treated pseudo word items mirrored the Thanks to Tal Schwarzmann, Katrina Wu, Henna Chaudhry,
results for treatment stimuli in all participants; however, the and Nicole Willcox for assistance with data collection and the
changes were less pronounced. Generalization to connected four children and their families.
speech was also observed with change over time in measures
of stress pattern match and PVC. Prior to treatment all
Declaration of interest
children had prosodic impairments in connected speech in the
moderate to severe range. Although connected PCC measures The authors are not aware of any conflicts of interest
were mostly mild to moderate prior to treatment, in combin- associated with this paper. The authors are solely responsible
ation with poor prosody, this resulted in a high level of for the content and writing of this paper. Parts of this study
perceived unintelligibility. The connected speech generaliza- were presented at the 2010 Conference on Motor Speech in
tion reported here is a highly desired outcome, as an increase Savannah, GA, USA, and the 2010 Speech Pathology
in matching prosodic patterns will lead to increased speech Australia National Conference in Melbourne, VIC, Australia.
intelligibility [58]. Additionally, generalization to connected
speech suggests that the use of orthographically biased pseudo References
word stimuli may promote learning during treatment and
1. Shriberg LD, Green JR, Campbell TF, McSweeny JL, Scheer AR. A
promote generalization to more complex untreated but diagnostic marker for childhood apraxia of speech: The coefficient
desirable behaviors. This suggestion requires further investi- of variation ratio. Clinical Linguistics & Phonetics 2003;
gation but such generalization from pseudo words has been 17(7):575–595.
2. Sussman HM, Marquardt TP, Doyle J. An acoustic analysis of
reported for other speech disorders in children [42].
phonemic integrity and contrastiveness in developmental apraxia of
speech. Journal of Medical Speech Language Pathology 2000;
Limitations 8(4):301–313.
3. Maassen B. Issues contrasting adult acquired versus developmental
The design of the current study limits the generalization of apraxia of speech. Seminars in Speech & Language 2002;
these results but provides preliminary evidence of treatment 23(4):257–266.
efficacy. When combined with the extant Ballard et al. [26] 4. Stackhouse J. Developmental verbal dyspraxia I: A review and
critique. European Journal of Disorders of Communication 1992;
and Van Rees et al. [30] studies, the findings reported here
27(1):19–34.
suggest that ReST intervention is a promising treatment 5. Thoonen G, Maassen B, Wit J, Gabreëls F, Schreuder R. The
worthy of further investigation. integrated use of maximum performance tasks in differential
A number of additional constraints on the generalizability diagnostic evaluations among children with motor speech disorders.
Clinical Linguistics & Phonetics 1996;10(4):311–336.
of the research exist. Each child was treated by a different 6. Ballard KJ, Djaja D, Arciuli J, James DG, van Doorn JL.
clinician and so there may have been a confound of treating Developmental trajectory for production of prosody: Analysis of
clinician. However, participant–clinician pairs were randomly lexical stress contrastivity in children aged 3 to 7 years and adults.
DOI: 10.3109/17518423.2014.906002 Prosodic treatment of CAS 145
Journal of Speech Language and Hearing Research 2012; 29. van Rees LJ, Ballard KJ, McCabe P, Macdonald-D’Silva AG,
55(6):1822–1835. Arciuli J. Training production of lexical stress in typically
7. James DGH. Hippopotamus is so hard to say: Children’s acquisition developing children with orthographically biased stimuli and
of polysyllabic words. Sydney, Australia: The University of Sydney; principles of motor learning. American Journal of Speech-
2006. Language Pathology 2012;21(3):197–206.
8. Arciuli J, Monaghan P, Seva N. Learning to assign lexical stress 30. Maas E, Barlow J, Robin D, Shapiro L. Treatment of sound errors
during reading aloud: Corpus, behavioral, and computational in aphasia and apraxia of speech: Effects of phonological
investigations. Journal of Memory and Language 2010; complexity. Aphasiology 2002;16(4/5/6):609–622.
63(2):180–196. 31. Semel E, Wiig E, Secord W. Clinical evaluation of language
9. Hall PK, Jordan LS, Robin DA. Developmental apraxia of speech: fundamentals – 4th ed. Australian Standardised Edition (CELF-4
Theory and clinical practice. Austin, TX: Pro-Ed; 1993. Australian). Pearson Inc.; 2006.
10. Davis BL, Jakielski KJ, Marquardt TP. Developmental apraxia of 32. Dunn LM, Dunn DM. Peabody picture vocabulary test.
speech: Determiners of differential diagnosis. Clinical Linguistics Minneapolis, MN: Pearson; 2007.
& Phonetics 1998;12(1):25–45. 33. American Speech-Language-Hearing Association. Guidelines for
11. Hickman LA. The apraxia profile. San Antonio, TX: The manual pure tone threshold audiometry. ASHA 1978;20:297–301.
Psychological Corporation; 1997. 34. Robbins J, Klee T. Clinical assessment of oropharyngeal motor
12. Shriberg LD, Aram DM, Kwiatkowski J. Developmental apraxia of development in young children. Journal of Speech and Hearing
speech: II. Toward a diagnostic marker. Journal of Speech, Disorders 1987;52:217–277.
Language, and Hearing Research 1997;40(2):286–312. 35. McLeod S. Sampling consonant clusters: Four procedures designed
13. Odell KH, Shriberg LD. Prosody-voice characteristics of children for Australian children. Australian Communication Quarterly 1997;
and adults with apraxia of speech. Clinical Linguistics & Phonetics Autumn:9–12.
2001;15(4):275–307. 36. Long SH, Fey ME, Channell RW. Computerized profiling 9.7.0.
14. Munson B, Bjorum EM, Windsor J. Acoustic and perceptual Cleveland, OH: Case Western Reserve University; 2006.
correlates of stress in nonwords produced by children with 37. American Speech-Language-Hearing Association. Childhood
suspected developmental apraxia of speech and children with Apraxia of Speech [Technical Report]. Available from: www.
phonological disorder. Journal of Speech, Language, and Hearing asha.org/policy. 2007.
Research 2003;46:189–202. 38. Dodd B, Hua Z, Crosbie S, Holm A, Ozanne A. Diagnostic
15. Guyette TW, Diedrich WM. A critical review of developmental Evaluation of Articulation and Phonology (DEAP). London,
apraxia of speech. In: Lass N, editor. Speech and language: England: The Psychological Corporation; 2002.
Advances in basic research and practice. Vol. 5. New York: 39. Neale MD. Neale analysis of reading ability. Melbourne, Australia:
Academic Press; 1981. pp 1–49. ACER Press; 1999.
16. Souza TN, Payão MC, Costa RC. Childhood speech apraxia in 40. Woodcock RW. Woodcock Reading Mastery Tests-Revised/
focus: Theoretical perspectives and present tendencies. Pró-fono: Normative Update (WRMT-R/NU). Allen, TX: Pearson
Revista de atualização cientı́fica 2009;21(1):76–80. Assessments; 1998.
17. Williams P, Stephens H. Nuffield Centre Dyspraxia Programme. 41. Wagner R, Torgesen JK, Rashotte CA. Comprehensive test of
Windsor, United Kingdom: The Miracle Factory; 2004. phonological processing (CTOPP). Austin, TX: Pro-Ed; 1999.
18. Arciuli J, Slowiaczek LM. The where and when of linguistic word- 42. Gierut JA, Morrisette ML, Ziemer SM. Nonwords and generaliza-
level prosody. Neuropsychologia 2007;45(11):2638–2642. tion in children with phonological disorders. American Journal of
19. Arciuli J, Cupples L. Effects of stress typicality during spoken word Speech-Language Pathology 2010;19(2):167–177.
recognition by native and nonnative speakers of English: Evidence 43. Gupta P, Lipinski J, Abbs B, Lin P-H, Aktunc E, Ludden D, Martin
from onset gating. Memory and Cognition 2004;32(1):21–30. N, Newman R. Space aliens and nonwords: Stimuli for investigating
20. Arciuli J, Cupples L. Effects of stress typicality during speeded the learning of novel word-meaning pairs. Behavior Research
grammatical classification. Language and Speech 2003; Methods, Instruments, & Computers 2004;36(4):599–603.
46(4):353–374. 44. Adobe Systems Incorporated. Adobe Audition 1.0 2003.
21. Slowiaczek LM, Soltano EG, Bernstein HL. Lexical and metrical 45. Deliyski DD, Shaw HS, Evans MK. Influence of sampling rate on
stress in word recognition: Lexical or pre-lexical influences? accuracy and reliability of acoustic voice analysis. Logopedics,
Journal of Psycholinguistic Research 2006;35(6):491–512. Phoniatrics, Vocology 2005;30:55–62.
22. Strand EA, Debertine P. The efficacy of integral stimulation 46. Murray E, McCabe P, Ballard KJ. A comparison of two treatments
intervention with developmental apraxia of speech. Journal of for childhood apraxia of speech: Methods and treatment protocol
Medical Speech Language Pathology 2000;8(4):295–300. for a parallel group randomised control trial. BMC Pediatrics 2012;
23. Strand EA, Skinder A. Treatment of developmental apraxia of 12:112.
speech: Integral stimulation methods. In: Caruso A, Strand E, 47. Maas E, Robin DA, Austermann Hula SN, Freedman SE, Wulf G,
editors. Clinical management of motor speech disorders in children. Ballard KJ, Schmidt RA. Principles of motor learning in treatment
New York: Thieme Medical Publishers Inc.; 1999. of motor speech disorders. American Journal of Speech-Language
24. Strand EA, Stoeckel R, Baas BS. Treatment of severe childhood Pathology 2008;17(3):277–298.
apraxia of speech: A treatment efficacy study. Journal of Medical 48. McCabe P, Ballard KJ. An Innovative Syllable Transition
Speech-Language Pathology 2006;14(4):297–307. Treatment Trial for Childhood Apraxia of Speech. Motor Speech
25. Ballard KJ, Robin DA, McCabe P, McDonald J. A treatment for Conference. Monterey, CA, USA; 2008.
dysprosody in childhood apraxia of speech. Journal of Speech, 49. Scruggs TE, Mastropieri MA. Summarizing single-subject
Language, and Hearing Research 2010;53(5):1227–1245. research: Issues and applications. Behavior Modification 1998;
26. Arciuli J, Cupples L. The processing of lexical stress during visual 22(3):221–242.
word recognition: Typicality effects and orthographic correlates. 50. Ruggero L, McCabe P, Ballard KJ, Munro N. Paediatric speech-
Quarterly Journal of Experimental Psychology 2006; language pathology service delivery: An exploratory survey of
59(5):920–948. Australian parents. International Journal of Speech-Language
27. Ševa N, Monaghan P, Arciuli J. Stressing what is important: Pathology 2012;14(4):338–350.
Orthographic cues and lexical stress assignment. Journal of 51. Thomas D, McCabe P, Ballard KJ. Parent training for rapid syllable
Neurolinguistics 2009;22(3):237–249. transitions treatment for childhood apraxia of speech: Fidelity of
28. Arciuli J, Cupples L. Would you rather ‘embert a cudsert’ or parent conducted treatment. Motor Speech Symposium. Sarasota,
‘cudsert an embert’? How spelling patterns at the beginning of FL, USA; 2014.
English disyllables can cue grammatical category. In: Khlentzos 52. Gillon GT, Moriarty BC. Childhood apraxia of speech: Children at
ACSD, editor. Mental states, Vol. 2: Language and cognitive risk for persistent reading and spelling disorder. Seminars in Speech
structure, Studies in language companion series. Amsterdam, and Language 2007;28(1):48–57.
Netherlands: John Benjamins Publishing Company; 2007. 53. McNeill BC, Gillon GT, Dodd B. Phonological awareness and early
pp 213–238. reading development in childhood apraxia of speech (CAS).
146 P. McCabe et al. Dev Neurorehabil, 2014; 17(2): 137–146

International Journal of Language & Communication Disorders graphic and perceptual analysis. International Journal of Speech-
2009;44(2):175–192. Language Pathology 2007;9(1):20–35.
54. Mauszycki SC, Wambaugh JL. The effects of rate control treatment 57. James DGH, van Doorn J, McLeod S. The contribution of
on consonant production accuracy in mild apraxia of speech. polysyllabic words in clinical decision making about children’s
Aphasiology 2008;22(7-8):906–920. speech. Clinical Linguistics & Phonetics 2008;22(4 and
55. Brendel B, Ziegler W. Effectiveness of metrical pacing in the 5):345–353.
treatment of apraxia of speech. Aphasiology 2007;22(1):77–102. 58. Klopfenstein M. Interaction between prosody and intelligibility.
56. Howard S. The interplay between articulation and prosody in International Journal of Speech-Language Pathology 2009;
children with impaired speech: Observations from electropalato- 11(4):326–331.

Appendix

Table A1. Treatment and non-treatment stimuli set with treatment words in bold (based on Arciuli et al. [1] and used in
Van Rees et al. [2]).

Phonemic transcription and Phonemic transcription and

SW orthography production used for imitation WS production used for imitation
coddol /kodFl/ adibe /Fdib/
combol /kombFl/ adoon /Fdun/
condan /kondFn/ amade /FmeId/
condey /kondF/ apoon /Fpun/
copet /kopFt/ bemade /bFmeId/
covan /kovFn/ bemiss /bFmIs/
mandan /mændFn/ bepade /bFpeId/
mapet /mæpFt/ beribe /bFraIb/
mappol /mæpFl/ bevade /bFveId/
maran /mæ7Fn/ abade /FbeId/
combet /kombFt/ amiss /FmIs/
conol /konFl/ bedibe /bFdaIb/
mambey /mæmbF/ bediss /bFdIs/
mandol /mændFl bedoon /bFdun/
manet /mænFt/ bemoon /bFmun/

References
[1] Arciuli J, Monaghan P, Seva N. Learning to assign lexical stress during reading aloud: Corpus, behavioral, and
computational investigations. Journal of Memory and Language 2010;63(2):180-196.
[2] van Rees LJ, Ballard KJ, McCabe P, Macdonald-D’Silva AG, Arciuli J. Training production of lexical stress in
typically developing children with orthographically biased stimuli and principles of motor learning. American Journal
of Speech-Language Pathology 2012:1058-0360_2012_11-0008.
Copyright of Developmental Neurorehabilitation is the property of Taylor & Francis Ltd and
its content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email
articles for individual use.